Re: [PATCH v4 i-g-t] tests/xe_gt_freq: Avoid RPe usage in subtests

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 




On 7/30/2024 10:07 PM, Riana Tauro wrote:
Hi Vinay

On 7/31/2024 4:31 AM, Vinay Belgaumkar wrote:
We are seeing several instances where the RPe, which can be altered by
pcode dynamically, is causing subtests to fail randomly. Instead of relying on it, we can use a mid frequency value for these subtests and avoid these
failures.

v2: Fix bug in the tolerance function. Remove rpe usage from freq_range*
as well (Badal). Fix test documentation to reflect change (Riana).

v3: Actual frequency cannot be guaranteed to follow the requested value

v4: Handle cases where RPe can be greater than min/cur freq.

Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2200
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2196
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2262
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2256
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2203
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2412
Closes?

Cc: Riana Tauro <riana.tauro@xxxxxxxxx>
Cc: Badal Nilawar <badal.nilawar@xxxxxxxxx>
Signed-off-by: Vinay Belgaumkar <vinay.belgaumkar@xxxxxxxxx>
---
  tests/intel/xe_gt_freq.c | 57 ++++++++++++++++++++++++++++------------
  1 file changed, 40 insertions(+), 17 deletions(-)

diff --git a/tests/intel/xe_gt_freq.c b/tests/intel/xe_gt_freq.c
index 93ebb5ed0..c9d59508b 100644
--- a/tests/intel/xe_gt_freq.c
+++ b/tests/intel/xe_gt_freq.c
@@ -26,6 +26,9 @@
  #include <sys/time.h>
    #define MAX_N_EXEC_QUEUES 16
+#define GT_FREQUENCY_MULTIPLIER    50
+#define GT_FREQUENCY_SCALER    3
+#define FREQ_UNIT_MHZ DIV_ROUND_CLOSEST(GT_FREQUENCY_MULTIPLIER, GT_FREQUENCY_SCALER)
    /*
   * Too many intermediate components and steps before freq is adjusted
@@ -70,9 +73,14 @@ static uint32_t get_freq(int fd, int gt_id, const char *freq_name)
      return freq;
  }
  -static uint32_t rpe(int fd, int gt_id)
+static bool within_expected_range(uint32_t freq, uint32_t val)
  {
-    return get_freq(fd, gt_id, "rpe");
+    /*
+     * GT Frequencies are requested at units of 16.66 Mhz, so allow
+     * that tolerance.
+     */
+    return (freq <= val + FREQ_UNIT_MHZ) &&
+        (freq >= val - FREQ_UNIT_MHZ);
  }
    static uint32_t get_throttle(int fd, int gt_id, const char *throttle_file)
@@ -128,6 +136,8 @@ static void test_freq_basic_api(int fd, int gt_id)
  {
      uint32_t rpn = get_freq(fd, gt_id, "rpn");
      uint32_t rp0 = get_freq(fd, gt_id, "rp0");
+    uint32_t rpmid = (rp0 + rpn) / 2;
+    uint32_t min_freq, max_freq;
        /*
       * Negative bound tests
@@ -142,16 +152,19 @@ static void test_freq_basic_api(int fd, int gt_id)
      /* Assert min requests are respected from rp0 to rpn */
      igt_assert(set_freq(fd, gt_id, "min", rp0) > 0);
      igt_assert(get_freq(fd, gt_id, "min") == rp0);
-    igt_assert(set_freq(fd, gt_id, "min", rpe(fd, gt_id)) > 0);
-    igt_assert(get_freq(fd, gt_id, "min") == rpe(fd, gt_id));
+    igt_assert(set_freq(fd, gt_id, "min", rpmid) > 0);
+    min_freq = get_freq(fd, gt_id, "min");
+    /* SLPC can set min higher than rpmid - as it follows RPe */
+    igt_assert(min_freq >= (rpmid - FREQ_UNIT_MHZ));
      igt_assert(set_freq(fd, gt_id, "min", rpn) > 0);
      igt_assert(get_freq(fd, gt_id, "min") == rpn);
        /* Assert max requests are respected from rpn to rp0 */
      igt_assert(set_freq(fd, gt_id, "max", rpn) > 0);
      igt_assert(get_freq(fd, gt_id, "max") == rpn);
-    igt_assert(set_freq(fd, gt_id, "max", rpe(fd, gt_id)) > 0);
-    igt_assert(get_freq(fd, gt_id, "max") == rpe(fd, gt_id));
+    igt_assert(set_freq(fd, gt_id, "max", rpmid) > 0);
+    max_freq = get_freq(fd, gt_id, "max");
+    igt_assert(within_expected_range(max_freq, rpmid));
      igt_assert(set_freq(fd, gt_id, "max", rp0) > 0);
      igt_assert(get_freq(fd, gt_id, "max") == rp0);
  }
@@ -168,13 +181,16 @@ static void test_freq_fixed(int fd, int gt_id, bool gt_idle)
  {
      uint32_t rpn = get_freq(fd, gt_id, "rpn");
      uint32_t rp0 = get_freq(fd, gt_id, "rp0");
+    uint32_t rpmid = (rp0 + rpn) / 2;
+    uint32_t cur_freq, act_freq;
        igt_debug("Starting testing fixed request\n");
        /*
       * For Fixed freq we need to set both min and max to the desired value        * Then we check if hardware is actually operating at the desired freq -     * And let's do this for all the 3 known Render Performance (RP) values. +     * And let's do this for all the 2 known Render Performance (RP) values
+     * RP0 and RPn and something in between.
       */
      igt_assert(set_freq(fd, gt_id, "min", rpn) > 0);
      igt_assert(set_freq(fd, gt_id, "max", rpn) > 0);
@@ -190,17 +206,20 @@ static void test_freq_fixed(int fd, int gt_id, bool gt_idle)
          igt_assert(get_freq(fd, gt_id, "act") == rpn);
      }
  -    igt_assert(set_freq(fd, gt_id, "min", rpe(fd, gt_id)) > 0);
-    igt_assert(set_freq(fd, gt_id, "max", rpe(fd, gt_id)) > 0);
+    igt_assert(set_freq(fd, gt_id, "min", rpmid) > 0);
+    igt_assert(set_freq(fd, gt_id, "max", rpmid) > 0);
      usleep(ACT_FREQ_LATENCY_US);
-    igt_assert(get_freq(fd, gt_id, "cur") == rpe(fd, gt_id));
+    cur_freq = get_freq(fd, gt_id, "cur");
+    /* If rpmid is around RPe, we could see SLPC follow it */
+    igt_assert(cur_freq >= (rpmid - FREQ_UNIT_MHZ));
        if (gt_idle) {
          igt_assert_f(igt_wait(xe_is_gt_in_c6(fd, gt_id), 1000, 10),
                   "GT %d should be in C6\n", gt_id);
          igt_assert(get_freq(fd, gt_id, "act") == 0);
      } else {
-        igt_assert(get_freq(fd, gt_id, "act") == rpe(fd, gt_id));
+        act_freq = get_freq(fd, gt_id, "act");
+        igt_assert(act_freq <= cur_freq + FREQ_UNIT_MHZ);
Can actual frequency be below the expected range of requested?

Yes, pcode can throttle it based on thermals.

Thanks,

Vinay.


Thanks,
Riana
      }
        igt_assert(set_freq(fd, gt_id, "min", rp0) > 0);
@@ -232,15 +251,17 @@ static void test_freq_fixed(int fd, int gt_id, bool gt_idle)
  static void test_freq_range(int fd, int gt_id, bool gt_idle)
  {
      uint32_t rpn = get_freq(fd, gt_id, "rpn");
+    uint32_t rp0 = get_freq(fd, gt_id, "rp0");
+    uint32_t rpmid = (rp0 + rpn) / 2;
      uint32_t cur, act;
        igt_debug("Starting testing range request\n");
        igt_assert(set_freq(fd, gt_id, "min", rpn) > 0);
-    igt_assert(set_freq(fd, gt_id, "max", rpe(fd, gt_id)) > 0);
+    igt_assert(set_freq(fd, gt_id, "max", rpmid) > 0);
      usleep(ACT_FREQ_LATENCY_US);
      cur = get_freq(fd, gt_id, "cur");
-    igt_assert(rpn <= cur && cur <= rpe(fd, gt_id));
+    igt_assert(rpn <= cur && cur <= rpmid + FREQ_UNIT_MHZ);
        if (gt_idle) {
          igt_assert_f(igt_wait(xe_is_gt_in_c6(fd, gt_id), 1000, 10),
@@ -248,7 +269,7 @@ static void test_freq_range(int fd, int gt_id, bool gt_idle)
          igt_assert(get_freq(fd, gt_id, "act") == 0);
      } else {
          act = get_freq(fd, gt_id, "act");
-        igt_assert(rpn <= act && act <= rpe(fd, gt_id));
+        igt_assert((rpn <= act) && (act <= cur + FREQ_UNIT_MHZ));
      }
        igt_debug("Finished testing range request\n");
@@ -262,17 +283,19 @@ static void test_freq_range(int fd, int gt_id, bool gt_idle)
  static void test_freq_low_max(int fd, int gt_id)
  {
      uint32_t rpn = get_freq(fd, gt_id, "rpn");
+    uint32_t rp0 = get_freq(fd, gt_id, "rp0");
+    uint32_t rpmid = (rp0 + rpn) / 2;
        /*
       *  When max request < min request, max is ignored and min works like
       * a fixed one. Let's assert this assumption
       */
-    igt_assert(set_freq(fd, gt_id, "min", rpe(fd, gt_id)) > 0);
+    igt_assert(set_freq(fd, gt_id, "min", rpmid) > 0);
      igt_assert(set_freq(fd, gt_id, "max", rpn) > 0);
      usleep(ACT_FREQ_LATENCY_US);
  -    /* Refresh value of rpe, pcode could have adjusted it */
-    igt_assert(get_freq(fd, gt_id, "cur") == rpe(fd, gt_id));
+    /* Cur freq will follow RPe, which could be higher than min freq */
+    igt_assert(get_freq(fd, gt_id, "cur") >= (rpmid - FREQ_UNIT_MHZ));
  }
    /**



[Index of Archives]     [AMD Graphics]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux